Original citation: Clough, A. R. and Edwards, R. S. (Rachel S.). (2015) Characterisation of hidden defects using the near-field ultrasonic enhancement of Lamb waves. Ultrasonics, 59 . pp. 64-71.
A note on versions: The version presented here may differ from the published version or, version of record, if you wish to cite this item you are advised to consult the publisher’s version. Please see the ‘permanent WRAP url’ above for details on accessing the published version and note that access may require a subscription. For more information, please contact the WRAP Team at: [email protected]
Characterisation of hidden defects using the near-field ultrasonic enhancement ofLamb waves
A.R. Clough, R.S. Edwards
Department of Physics, University of Warwick, Coventry, CV4 7AL
Abstract
Defects that propagate from the inside of a structure can be difficult to detect by traditional non-destructive inspectionmethods. A non-contact inspection method is presented here that uses the near-field interactions of ultrasonic Lambwaves to detect defects propagating into a 1.5 mm thick aluminium sheet from the opposite side to that which isinspected. Near-field interactions of the S0 Lamb waves with the defects are shown to give rise to a characteristicincrease in the wave magnitude, which is used to position and characterise these hidden defects. It is shown that suchdefects are difficult to detect from a study of their influence on ultrasonic transmission alone. Single defects of differentdepths, and systems of multiple defects with varying separations and relative depths, are successfully detected in bothexperimental trials and FEM simulations. Reliable single defect detection is achieved for defects with a minimumdepth of 30% of the plate thickness, and resolution of multiple defects is achieved for defect separations of 5 mm.
Early detection of cracking in industrial applicationsallows replacement of the faulty part, preventing com-ponent failures which are costly both in economic andenvironmental terms [1]. Surface-breaking cracks, suchas defects caused by stress-corrosion cracking (SCC), inwhich defect growth occurs when a material is placedunder stress in a corrosive environment, are of con-cern in industrial pipework and chemical storage sys-tems [1–4]. SCC defects typically have a size scale ofseveral millimetres with a complicated branched struc-ture, and can occur singly or in groups of defects locatedclose together [1].
Such defects are traditionally detected using dye pen-etrant inspection, however, the application of this tech-nique requires extended downtime of the system un-der test as the inspection cannot be done during oper-ation [5,6]. Dye penetrant inspection also requires directaccess to the damaged surface and so success is limitedwhen access to the object under test is restricted. Radio-graphic inspection can also be employed to detect thesetypes of defects, however, safety concerns arising fromthe use of ionising radiation can limit the application inindustrial settings [7].
Ultrasonic inspection is an attractive alternative todye penetrant inspection, however, conventional piezo-electric ultrasonic transducers require the use of cou-plant, and as such have a limited capacity for scan-ning large samples or at elevated temperatures [4,8–10].The use of non-contact ultrasonic generation and de-tection methods, such as laser ultrasonics [11] and elec-tromagnetic acoustic transducers (EMATs) [12] removesthe need for couplant, and thereby provides the poten-tial to perform simple scanning inspections on compo-nents [13].
The use of ultrasonic waves with displacementsthroughout the thickness of a material, such as Lambwaves in sheets and guided waves in pipes [14,15], enablesinspection of areas of the system that cannot be accesseddirectly, such as the internal surface of a pipe [8,16]. Sev-eral long-distance ultrasonic inspection methods existthat monitor in the defect far-field (defined as the regionstarting at a distance of several wavelengths away fromthe defect [17]) through changes in the reflection or trans-mission of guided waves as they interact with surface-breaking defects, and these methods are capable of esti-mating the position and depth of defects over a distanceof several metres [8,9]. However, the reflectivity of small,shallow defects is low, which restricts the size of defectsthat can be detected by this far-field approach. The am-
Preprint submitted to Elsevier January 20, 2015
plitude of the wave that is transmitted past a defect canalso be used to estimate depth; a larger reduction in am-plitude indicates a deeper defect [18]. However, whenthe defect is very small or not orientated perpendicularto the direction of wave propagation, detection can bedifficult to achieve for a detector located far away fromthe defect, as diffraction around the defect can preventchanges in transmission from being observed [8].
In addition, SCC defects are not a singular occurrenceand often multiple defects occur in a component, eachof which may grow to a different depth [1]. Interactionof ultrasonic guided waves with these clusters of de-fects offers significant complexity to the measurements,for example due to the reduced amplitude for reflectedwaves from later defects. Interactions therefore tend tobe dominated by the relative depth of the defect that isfirst encountered by the incident wave, and the defectseparation [19], with overall transmission tending to bedominated by the deepest defect [18].
An alternative to these far-field studies is the use ofa scanning inspection system to monitor changes in theultrasonic propagation as and when they occur, therebyminimising the influence of issues such as diffractionaround small defects or low defect reflectivities [20–24].The most distinctive near-field effect is the enhancementof an ultrasonic surface wave, which is observed whenan ultrasonic source or detector is scanned directly overa defect (figure 1).
1.1. Near-field enhancementsThe enhancement of surface waves has been observed
for Rayleigh [20,21] and Lamb waves [22,23], for surface-breaking defects which propagate into the sample fromthe side which is inspected. When a detector is scannedover such a defect, constructive interference betweenthe incident wave mode (Rayleigh or Lamb) and wavemodes that are reflected and mode converted at the de-fect gives rise to a characteristic increase in the signalamplitude, and an increase in the magnitude of individ-ual frequency components of the incident wave [20–22]. Asimilar mechanism is responsible for the enhancementwhen the ultrasonic source is passed over the defect,with additional contributions when using laser ultrason-ics arising from changes in the boundary conditions andspatial profile of the laser generation source at the de-fect [21,23].
When using Lamb waves, mode conversion can oc-cur between all of the wave modes that are supported atthe frequency-thickness of inspection, with more modesavailable at higher frequency-thicknesses [22,25]. Follow-ing interaction, these waves propagate away from thedefect with velocities determined by the mode (shown
(a)
Ultrasonic source
Hidden defect
Component under test
(b)
Figure 1: A cut-through of a sample showing scanninglaser detection of the defect near-field in a sheet for de-fects on the same side as the inspection (a) and hiddendefects (b).
in figure 2), and enhancement is only observed closeto the defect when these reflected and mode convertedwaves arrive within the same time window as the inci-dent mode [22].
The location at which the enhancement occurs can beused to determine the position of the defect [22,24]. In ad-dition, the magnitude of the enhancement has been usedto give an estimate of the severity of the defect, with alarger enhancement indicating a more severe defect dueto the larger reflections [22–24]. The remote, non-contactnature of laser ultrasonics and EMAT inspection allowsfor the scanning inspection that is essential for thesemeasurements, however, the high spatial resolution of-fered by laser ultrasonics makes it the ideal candidatefor near-field inspection [24].
Previous studies have examined the enhancement ofguided waves as the laser detector or source is passedover a surface-breaking defect that propagates into thesurface of the material from the inspection side (figure1a). However, many defects can propagate outwards
2
Figure 2: Lamb wave group velocity dispersion curvefor aluminium.
from the inside of a structure, such as a pipe or storagetank, and leave no visible indication of the defect on theouter surface of the structure; this situation is similarto that shown in figure 1b. The present study identi-fies near-field ultrasonic enhancements arising from de-fects that propagate into the material from the far side ofthe structure, relative to the testing surface, and demon-strates how these enable positioning and characterisa-tion of such hidden defects. The near-field interactionswith multiple hidden defects, where two closely-spaceddefects are present, are shown as the defect separationand the depths of the defects relative to one another arevaried.
2. Methods
Inspections were performed on 1.5 mm thick alu-minium plates (300 x 300 x 1.5 mm) into which ar-tificial defects with a v-shaped side profile (figure 3)were cut by laser micro-machining. Defects of length25 mm and of different depths, d, were produced (rang-ing from 5% ≤ d ≤ 100% of the through-thickness of thesheet) at a position such that reflections from the sheetedges were minimised. The v-shaped profile was cho-sen as a simplified representation of the opening partof a SCC defect, and the average defect opening widthwas 282 ± 16µm. Inspection in the near-field of the de-fect was carried out by scanning both the source anddetection lasers over the undamaged side of the platesuch that the detector passed over the defect region. Thescanning was performed using a linear stage with a scanstep of 0.05 mm.
Generation of ultrasound used a pulsed Q-switchedNd:YAG laser (1064 nm wavelength, 10 ns rise time),operating in the thermoelastic regime to limit sampledamage, focused to a spot with 2 mm diameter to give abroadband ultrasonic source with significant frequency
Figure 3: A schematic diagram showing scanning laserdetection inspection of laser micro-machined defects ofdepth d in 1.5 mm thick aluminium sheets.
content in the range of 0.1 - 6 MHz [11]. Both symmet-ric and antisymmetric Lamb waves were excited, withvelocities dependent on the frequency of the excitationand the thickness of the sheet [25]. Detection of ultra-sound was carried out using a two-wave mixing interfer-ometer (1550 nm wavelength) from Intelligent OpticalSystems (IOS) [26], which provides a measure of the out-of-plane displacement on the material surface. The IOSdetector has a spot size of 200 µm, allowing for highspatial resolution when scanning. The generation anddetection lasers were held at a fixed separation in orderto simplify the interpretation of the multi-modal ultra-sonic signals received, and to minimise the influence ofattenuation [22,23].
In addition to the experimental measurements, a se-ries of 3D finite element method simulations were car-ried out using PZFlex. Simulated v-shaped defects ofdifferent depths in 1.5 mm thick sheets were studied byusing a dipole force located a fixed distance away fromthe defect to simulate the laser generation. The out-of-plane displacement at the nodes on the undamaged sideof the sheet, for a line passing over the mid-point of thedefect, was recorded in a similar fashion to the experi-mental data. A fixed separation was not maintained inthe simulation as the calculation of new boundary con-ditions at each new source location is very time consum-ing. The element spacing in the FEM simulations was0.1 mm.
3. Results
3.1. Single defect
A-scans were recorded at each scan position and adja-cent scans were stacked to form B-scans, with an exam-ple B-scan shown in figure 4 and A-scans in the lowerpanels of figure 5. Negative detector position on the B-scan corresponds to both the detector and generator po-sitions being on the same side of the defect, prior to in-
3
Figure 4: A B-scan image for an experimental scanacross a 100% through-thickness defect located at0 mm. The time window used to monitor enhancementsis shown by the region bounded by dashed lines.
teraction (measuring incident and reflected waves); pos-itive detector positions here correspond to the generatorand detector being on opposite sides of the defect (mea-suring transmission). The scan was performed perpen-dicular to the defect, as shown in figure 3, to simplifythe analysis.
For the frequency-thickness region of interest, theonly supported wavemodes are the S0 and A0 waves.These are broadband with a central frequency of1.05 MHz, and the exact arrival times depend on thevelocity for each frequency (figure 2) [25]. Incident S0and A0 modes have constant arrival time during the scan(figure 4). Reflected and mode converted waves have avarying arrival time, with several such waves visible onthe B-scan. When the incident S0 mode interacts withthe defect it will undergo several processes; reflection ofthe incident wave to produce a backwards travelling S0wave, transmission forwards under the defect, and modeconversion to other modes that are supported at the samefrequency-thickness, in this case an A0 wave, whichwill propagate both forwards and backwards from thedefect [22,28,29]. These interactions are shown schemati-cally in figure 6.
Full identification of the wavemodes present requiresuse of time-frequency techniques to identify overlap-ping modes. However, by taking the central frequencyas being dominant, an approximate identification of thewaves on figure 4 is possible. For the central fre-quency velocities are 5110.1 ms−1 for the A0 modeand 3160.9 ms−1 for the S0. The incident and trans-mitted S0 modes arrive at approximately 10 µs, withlow amplitude (these have predominantly in-plane mo-tion [25], while the detection system is sensitive to out-of-plane displacements). In the negative position re-gion, corresponding to incident and reflected waves, a
(a)
(b)
Figure 5: A-scans (lower panels) and time-frequencyrepresentations (upper panels) taken on a defect freesheet (a) and for a 75% through-thickness defect fora detector position of 0.05 mm prior to the defect (b),shown overlain with theoretical arrival times of Lambwaves. The region of interest of the S0 mode is high-lighted.
strong mode-converted S0-A0r mode is observed, ar-riving a time ∆tS 0−A0r = |x|(1/vA0 + 1/vS 0) after theincident S0 wave, where vA0 is the velocity of the A0mode, vS 0 is the velocity of the S0 wave, and x is thedetector position relative to the defect. A weaker S0-S0r is also observed, but with much smaller amplitudedue to being primarily in-plane. Following transmis-sion (positive detector positions) the S0-S0 transmittedmode continues to arrive at around 10 µs. However, thearrival time of the S0-A0 transmitted mode varies de-pending on the distance of the detector from the slot(and hence how far the wave propagates as an S0 andas an A0 wave), with the time-varying arrival time oc-curring a time ∆tS 0−A0t = x(1/vA0 − 1/vS 0) after the S0-S0 transmitted wave. The defect position is shown bythe point at which these reflected and mode-convertedwaves originate (detector position of 0 mm).
Individual Lamb modes and their frequency-dependent behaviour can be fully identified by usinga sonogram time-frequency representation, which is
4
Figure 6: Interactions of the incident S0 wave encoun-tering a hidden v-shaped defect. Subscript i correspondsto an incident wave, r is a reflected wave produced at thedefect, and t is a wave transmitted past the defect.
created by splitting the time-domain data into manyequal sized sections and producing a Fourier transformof each [27]. The theoretical arrival times of individualmodes are overlain onto the sonograms to enableidentification of modes, producing figures such asfigure 5 [22,23].
Initial investigations were carried out on sheets thatcontained a single isolated defect, with experiments andsimulations performed with the detector scanned acrossthe undamaged side of the sheet, as shown in figure 1b.Enhancements were observed at several frequencies andarrival times on the sonograms, corresponding to inter-actions of different modes with the defects; see, for ex-ample, the increased magnitude in the windowed regionof figure 5a compared to that on figure 5b. Enhance-ment of the S0 mode was chosen for full analysis as itcorresponds to the earliest arriving incident Lamb waveat the detector, arriving at approximately 10 µs in fig-ure 4, and it can, therefore, be studied without inter-ference from other modes in a defect free sheet. For asheet containing defects, any changes that occur in thisregion are caused by interactions between the incidentS0 wave and those waves reflected and mode convertedat the defect.
The maximum magnitude of the sonogram was mea-sured in a chosen region corresponding to the inci-dent S0 mode, at frequency-thicknesses between 0.75 -1.35 MHz.mm, and arrival times (for the fixed exper-imental source to detector separation) between 9.75 -11.6 µs. This region is shown by the dashed box onfigure 5. By tracking the peak magnitude in this regionduring a scan, variations in the chosen wavemode canbe identified. This variation in the magnitude has pre-viously been used to characterise the ‘amount’ of thatwave which is present at a given detector position [22–24].
(a)
(b)
Figure 7: Variation in the sonogram magnitude of the S0wave in the chosen region as a function of detector po-sition for experimental (a) and simulated (b) scans overa hidden defect with a depth of 50% of the sheet thick-ness.
An example of the variation of the magnitude during ascan for the chosen S0 wave, incident on a 50% through-thickness defect, is shown in figure 7 for experimental(a) and simulated (b) data.
In the experimental data (figure 7a), the left hand re-gion has a steady average magnitude, showing that themagnitude comes from the incident wave only. Theright hand region (where the source and detector areon opposite sides of the defect) shows the magnitudeof the S0 wave returning to the pre-defect level dueto diffraction underneath and around the defect, high-lighting the difficulty in using changes in transmissionto characterise small length defects when measuring inthe far-field. The model data shows a similar behaviour,with an overall downward slope due to the spread of thewave front as the separation of generation and detectionpoints increases.
When the detector is in the near-field of the defect,defined here as when the detector is within ±4 mm fromthe defect midpoint, interactions between reflected,transmitted and mode-converted waves occur. Betweendetector positions of -4 and 0 mm, the reflected and
5
mode converted waves arrive within the same time win-dow as the incident S0 wave, leading to wave superpo-sition and interference [22] (also shown on the B-scan,figure 4). In the region of the scan in which this inter-ference occurs, the reflected and mode converted waveschange arrival time and phase relative to the incident S0wave as the distance to the defect changes [28,29], withregions of constructive and destructive interference ob-served. This is shown by the distinctive pattern of mul-tiple enhancement peaks in figure 7. The enhanced sig-nals are over twice the magnitude of the incident S0wave and give a clear indication of the presence of adefect.
The interactions of the incident S0 mode with the de-fect are shown schematically in figure 6; this shows re-flection of the incident wave to produce the reflected(backwards travelling) S0 wave, a transmitted (forwardtravelling) wave, and mode-conversion to other modessupported at this frequency thickness, in this case an A0wave. This propagates both forwards and backwardsfrom the defect, with the arrival times dependent on thevelocities [22,25,28,29].
At certain points in the near-field, the phase differ-ence between incident and reflected / mode-convertedwaves is small, giving constructive interference and anincreased signal magnitude. At other points, the wavesare out-of-phase with the incident wave mode, such thatdestructive interference can occur; this produces thedrops in magnitude at a detector position of -1 mm infigure 7. The main enhancement peak (peak 1) is ob-served very close to 0 mm on the scan, where the modesinterfere constructively; the second peak is observed at-2 mm when the reflected and mode converted wavesagain come into phase with the incident mode [22,25].
When the detector moves past the defect the wavesthat are present in the time window analysed are thetransmitted S0 wave and the forward travelling modeconverted A0 wave [28,29]. These interfere construc-tively, as before, giving rise to the broader peak centredaround a detector position of 5 mm in figure 7a. Fol-lowing transmission there is only a slow change in therelative phase difference between the two modes, allow-ing them to stay in phase over a longer distance; there isno path difference between transmitted S0 and A0, andthe shape of the enhancement peak is due to the differentvelocities and phases of the modes.
To quantify the enhancement an enhancement factorE f is calculated [22,23], and is found for the enhancementpeaks shown at detector positions of -2 mm and 0 mm infigure 7. E f is given by the ratio of the enhanced magni-tude, EEnhanced (the peak magnitude), to the magnitudethat is present when there is no defect, ENoDe f ect (taken
(a)
(b)
Figure 8: Variation in the enhancement factor of the S0wave in the region of interest as a function of defectdepth, for experimental (a) and simulated (b) scans overa single hidden defect.
from the far-field incident wave for the experiments, andconsidering the attenuation for the models),
E f =EEnhanced
ENoDe f ect. (1)
The S0 enhancement factor is found to vary as a func-tion of defect depth and is shown in figure 8 for en-hancements measured at both positions. The enhance-ment factors increase with increasing defect depth forboth experimental (figure 8a) and simulated data (fig-ure 8b). This increasing enhancement factor is causedby the increase in the reflection coefficient of the S0wave and the increase in the amount of mode conversionto A0 waves experienced at deeper defects [28,29]. Theenhancements observed in the simulations are slightlylarger than their experimental counterparts as they arerecorded on a single point node, whereas the experimen-tal data is averaged over the 200 µm diameter detectorspot, but follow the same trend.
A measure of the severity of the hidden defect can beobtained from the magnitude of the enhancement factor.This can be assessed from either enhancement peak, as
6
Figure 9: Set-up for FEM simulations on componentscontaining multiple defects of fixed depth (50% of sheetthickness) separated by a distance x.
long as the choice of peak is consistent between inspec-tions, or by considering both together to allow for vari-ations in the reflection coefficient due to crack rough-ness. The pattern of the enhancement peaks shown infigure 7 is consistent between defects of different depth,and hence the hidden defect can be positioned by thelocation at which the central enhancement peak occurs.This is particularly important for the detection of hiddendefects where there is no visible sign of a defect on thematerial surface.
3.2. Multiple defects
In many occurrences of SCC the material will de-velop defects at multiple locations [1]. Each crack willcontribute to the failure of the component, and hencereliable detection and resolution of multiple hidden de-fects and a measure of their relative depths and positionsis of interest.
Investigation into the near-field interactions betweenan incident S0 wave and multiple hidden defects wascarried out using FEM simulations to allow investiga-tion of a wide range of defect combinations. Figure 9shows the simulation set-up used to investigate a pair ofdefects, both with a depth of 50% of the sheet thickness(1.5 mm), with separation x. This separation was variedbetween x = 0.2 mm (corresponding to when the twodefects are effectively conjoined) and 12 mm. The samedata processing procedure as for the single defects wasperformed, and the variation in the sonogram magnitudefor each scan is shown in figure 10.
For small separations between the two defects (lessthan 1 mm) the pattern of near-field enhancement of theS0 wave is very similar in form to that of the single de-fect, with the first visible differences arising at a sepa-ration of 1 mm. At this separation the overall enhance-ment shows interference between two sets of near-fieldenhancements, both of which resemble the single defect
(a)
(b)
Figure 11: Interactions of the incident S0 wave with twohidden v-shaped defects of 50% depth.
shape, with a small offset from one another. The inter-ference between these enhancements makes it difficultto determine the depths of the defects using the cali-bration obtained from the single defect measurements,however, it can be surmised from the pattern that thedefect is more complex than a single defect. As the de-fect separation increases it becomes possible to resolvetwo clear near-field enhancements, each one associatedwith a defect, and for the 50% defects used here twoseparate enhancements are clearly resolvable for defectseparations of 5 mm and larger.
The first enhancement follows the same mechanismas for a single defect. The second enhancement is cre-ated by the same interference mechanism, however, thewave incident on the second defect is comprised of thetransmitted S0 wave and the forward travelling com-ponent of the mode converted A0 wave from the firstdefect, shown in figure 11. This leads to an increasedcomplexity of the signal analysis when the defects areclose together and both of these wavemodes arrive in thetime window for analysis (i.e. the second enhancementlies on top of the peak following transmission, centredaround 5 mm on figure 7). When the detector passesover the second defect these additional mode conver-sions must be considered, providing additional contribu-tions to the enhancement [28,29], with interactions shownin figure 11b. These additional wave modes providevarying contributions to the enhancement as the sepa-ration increases, as the increased propagation distanceleads to a varying phase difference with respect to the
7
Figure 10: Variation in the sonogram magnitude of the studied S0 wave for two defects that are 50% of the through-thickness of the sheet, for increasing separation between the defects.
S0 wave of interest, meaning that at some positions theirinteractions will be more constructive in nature, and atother positions destructive. This will affect the analy-sis for defect separations of 7 mm or less (see velocitiesin figure 2 and the width of the enhancement peaks infigure 10).
For single defects the near-field enhancement varieswith defect depth, and this variation can also be ex-ploited in order to determine the depths of multipledefects relative to one another for certain separations.The near-field enhancement of the S0 wave was exam-ined for two different defect separations, x = 1 mm andx = 5 mm, for varying depths, giving poorly resolved(1 mm) and almost resolved defects (5 mm). One of thedefects was fixed at 50% of the sheet thickness and theother defect was varied in depth, between 10% and 90%of the sheet thickness, with the detection encounteringthe 50% defect first (figures 12a and 13a for 1 mm and5 mm separations respectively), or second (figures 12band 13b).
Figures 12a and 12b (separation of 1 mm) show smallvariations in the structure of the enhancement peakswhen compared to those for a single defect (figure 7b),with the behaviour dominated by the larger defect ineach case. Once the correct enhancement peaks areidentified, a general trend in the behaviour is again ob-served, with larger enhancements with increasing defectdepths.
The increasing enhancement with defect depth whenthe deeper defect is encountered first is due to the in-creased contribution to the enhancement from the A0r1and S0r1 waves (figure 11a) as the defect depth in-creases, as for the single defect case [22,28,29]. This leadsto a subsequent decrease in the magnitude of the S0t1
and A0t1 modes interacting with the second defect, re-ducing the enhancement that can be observed from thesecond defect.
When the smaller defect is encountered first, the am-plitude of the transmitted waves A0t1 and S0t1 can belarge, and therefore larger magnitude additional reflec-tions can be produced (e.g. S0t1-S0r2 and A0t1-A0r2)which contribute to the enhancement at the second de-fect. This leads to the large enhancement observed at thesecond defect position in figure 12 when the second de-fect is deeper than the first. The presence of the smallerinitial defect can still be observed by the deviations ofthe enhancement pattern from the single defect case. Toobtain a more complete idea of the defect depths rela-tive to one another a scan can be performed from bothdirections.
In figure 13, for 5 mm separation, it is possible todetermine two different enhancements for all permuta-tions of the two defects; note that the second enhance-ment occurs in addition to the transmission enhance-ment peak (centred around 5 mm on figure 7). The sepa-ration means that both transmitted modes will still affectthe overall enhancement at the second defect, however,both sets of enhancement peaks have the same structureas the single defect enhancements shown in figure 7. Anestimate of the defect positions can be made by the lo-cation at which the central peaks occur.
The enhancement factor for a single simulated 50%depth defect was found to be 1.72 (figure 8b). For thecases in figure 13a where the first defect encountered is50% of the sheet thickness, the enhancement from this50% defect was found to be consistent for the double de-fect geometry. However, the presence of the initial 50%defect had the effect of increasing the enhancement fac-
8
(a)
(b)
Figure 12: Variation in the sonogram magnitude of theS0 wave in the region of interest for two hidden defectsseparated by 1 mm when the detector first encounters afixed 50% depth defect followed by a defect of varieddepth (a), and when the detector encounters the fixeddefect second (b).
tors for the second, variable depth defects, due to thepresence of the transmission enhancement, with an in-crease in the enhancement factor from 1.10 to 1.41 fora 10% defect and 2.19 to 2.96 for a 90% defect. Thesignificance of this increased enhancement at the 5 mmseparation can be highlighted by comparing the 10%then 50% defect pairings from figures 12 and 13; thesmaller defect was undetectable for the 1 mm separa-tion, but clear for the larger separation.
When the 50% depth defect is the second defect en-countered the enhancement factor for this defect wasfound to increase to above the single defect value of1.72, for the same reason as the increase in enhancementdiscussed above. A larger increase was seen for a deeperinitial defect, with the 50% depth defect enhancementfactor increasing to 2.14 for a 10% depth first defect, to2.58 for a 90% depth first defect. Figure 13 thereforeshows that, although enhancement factors can be foundfor each defect at a separation of 5 mm, the presence ofmultiple defects can act to alter the magnitude of the ob-
(a)
(b)
Figure 13: Variation in the sonogram magnitude of theS0 wave in the region of interest for two hidden defectsseparated by 5 mm when the detector first encounters afixed 50% depth defect followed by a defect of varieddepth (a), and when the detector encounters the fixeddefect second (b).
served enhancements when compared to the single de-fect case. The nature of the changes to the enhancementare dependent upon which defect is encountered first,and so to obtain as much information about the defectsystem as possible the sample should be scanned fromboth directions. However, the enhancements can still beused as a guide to approximate the defect sizes, and forseparations of 7 mm or larger at this analysis frequencyrange, when the enhancement patterns are clearly sepa-rated, the sizing regains accuracy.
4. Conclusions
It has been shown that near-field ultrasonic enhance-ment of a fundamental symmetric Lamb wave incidenton a hidden defect can be used to position the defect andobtain an estimate of its severity. The successful iden-tification of multiple defects has been shown for twodefect systems for a variety of different separations be-tween the defects and for defects of different relative
9
depths. The near-field scanning technique is especiallyuseful for resolution of individual defects within a de-fect cluster, which can be difficult to achieve using far-field methods.
The detection of hidden defects by this method is ofparticular merit when compared to the traditional dye-penetrant approach, which can only be applied to an ac-cessible surface. Higher resolution would be possibleby using pulses of higher frequency and shorter timeduration, however, care must be taken to avoid increas-ing the complexity of the signals by allowing the pres-ence of more than just the fundamental S0 and A0 Lambwave modes. The near-field inspection method couldbe used in conjunction with a long range guided wavemethod, where the far-field inspection could be used toprovide the approximate location of a defect cluster be-fore the near-field inspection is used to quantify the na-ture of the defects within that cluster.
Acknowledgments
This work was funded by the European ResearchCouncil under grant 202735, NonContactUltrasonic,ERC Starting Independent Researcher Grant.
5. References
References
[1] R A Cottis. Stress corrosion cracking. NPL, 2000.[2] C Manfredi and J L Otegui. Failure by SCC in buried pipelines.
Engineering Failure Analysis, 9:pp. 495–509, 2002.[3] B. Y. Fang, A. Atrens, J. Q. Wang, E. H. Han, Z. Y. Zhu, and
W. Ke. Review of stress corrosion cracking of pipeline steels inlow and high ph solutions. Journal of Materials Science, 8:pp.127–132, 2003.
[4] S. G Polyakov and A. A. Rybakov. The main mechanisms ofstress corrosion cracking in natural gas trunk lines. Strength ofMaterials, 41:pp. 456–463, 2009.
[5] A. de Sterke. A practical introduction to penetrants. Nonde-structive Testing, 1:pp. 306–307, 1968.
[6] N. P. Migun, A. B. Gnusin, and I. V. Volovich. Some possibili-ties of enhancing the efficiency of penetrant inspection. Russianjournal of nondestructive testing, 41:pp. 452–455, 2005.
[7] A. Maksimenko, M. Ando, H. Sugiyama, and E. Hashimoto.Possibility of computed tomographic reconstruction of cracksfrom X-ray refraction contrast. Japanese Journal of AppliedPhyiscs, 44:pp. 633–635, 2005.
[8] M. J. S. Lowe, D. N. Alleyne, and P. Cawley. Defect detection inpipes using guided waves. Ultrasonics, 36:pp. 147–154, 1998.
[9] P. Cawley and D. N. Alleyne. The use of Lamb waves for thelong range inspection of large structures. Ultrasonics, 34:pp.287–290, 1996.
[10] M. Kobayashi and C-K. Jen. Piezoelectric thick bismuth ti-tanate/lead zirconate titanate composite film transducers forsmart NDE of metals. Smart Materials and Structures, 13:pp.951–956, 2004.
[11] C.B. Scruby and L.E. Drain. Laser ultrasonics. Techniques andapplications. Adam Hilger, 1990.
[12] M Hirao and H Ogi. EMATs for science and industry. Noncon-tacting ultrasonic measurements. Kluwer Academic Publishers,2003.
[13] R. J. Dewhurst, C. Edwards, A. D. W. McKie, and S. B. Palmer.A remote laser system for ultrasonic velocity measurement athigh temperatures. Journal of Applied Physics, 63:pp. 1225–1227, 1988.
[14] M. Castaings and P. Cawley. The generation, propagation, anddetection of Lamb waves in plates using air-coupled ultrasonictransducers. Journal of the Acoustical Society of America, 100:pp. 3070–3077, 1996.
[15] T. Hayashi and M. Murase. Defect imaging with guided wavesin a pipe. Journal of the Acoustical Society of America, 117:pp.2134–2140, 2005.
[16] B. Massarey, C. Raemy, and P. Fromme. High-frequency guidedultrasonic waves for hidden defect detection in multi-layered air-craft structures. Ultrasonics, 54(7):pp. 1720–1728, 2014.
[17] J. David and N. Cheeke. Fundamentals and applications of ul-trasonic waves. CRC Press, 2002.
[18] R. S. Edwards, S. Dixon, and X. Jian. Characterisation of de-fects in the railhead using ultrasonic surface waves. NDT & EInternationa, 39(6):pp. 469–475, 2006.
[19] A. Lovstad and P. Cawley. The reflection of the fundamental tor-sional guided wave from multiple circular holes in pipes. NDT& E International, 44:pp. 553–562, 2011.
[20] R. S. Edwards, S. Dixon, and X. Jian. Enhancement of theRayleigh wave signal at surface defects. Journal of Physics D:Applied Physics, 37:pp. 2291–2297, 2004.
[21] S. Dixon, B. Cann, D. L. Carroll, Y. Fan, and R. S. Edwards.Non-linear enhancement of laser generated Rayleigh waves bycracks. Nondestrutive Testing and Evaluation, 23:pp. 25–34,2008.
[22] A. R. Clough and R. S. Edwards. Lamb wave near field enhance-ments for surface-breaking defects in plates. Journal of AppliedPhysics, 111:p. 104906, 2012.
[23] A. R. Clough and R. S. Edwards. Scanning laser source Lambwave enhancements for defect characterisation. NDT & E Inter-national, 62:pp. 99–105, 2014.
[24] F. Hernandez-Valle, A. R. Clough, and R. S. Edwards. Stresscorrosion cracking detection using laser/laser and laser/EMATtechniques. Corrosion Science, 78:pp. 335–342, 2014.
[25] J. L. Rose. Ultrasonic waves in solid media. Cambridge Uni-versity Press, 1999.
[26] M. B. Klein, G. D. Bacher, A. Grunnet-Jepsen, D. Wright, andW. E. Moerner. Homodyne detection of ultrasonic surface dis-placements using two-wave mixing in photorefractive polymers.Optics Communications, 162:pp. 79–84, 1999.
[27] M. Niethammer, L. J. Jacobs, J. Qu, and J. Jarzynski. Time-frequency representation of Lamb waves. Journal of the Acous-tical Society of America, 109:pp. 1841–1847, 2001.
[28] B. Kim and Y. Roh. Simple expression of the reflection andtransmission coefficients of fundamental Lamb waves by a rect-angular notch. Ultrasonics, 51:pp. 734–744, 2011.
[29] M. Castaings, E. Le Clezio, and B. Hosten. Modal decompo-sition method for modeling the interaction of Lamb waves withcracks. Journal of the Acoustical Society of America, 112:pp.2567–2582, 2002.
